Re: [Quantlib-users] Corporate bond analytics approach

[<mat...@gm...>]

While I am probably not the best person to give advice on this matter, here is something to potentially add to the discussion:

 

Choice between approach 1 or 2:

Maybe this can be depending on the bond/issuer rating. For IG, discounting with spreaded curves may be okay while for HY you want to use a term structure implied from CDS.

This question reminds me of https://arxiv.org/abs/0912.4623 - a dated paper, but could be still useful?

 

Implementation:
Maybe you can (re-) use ORE’s QuantLib extension which brings this one: https://github.com/OpenSourceRisk/Engine/blob/master/QuantExt/qle/pricingengines/discountingriskybondengine.cpp

 

Pros/Cons:

Depending on your portfolio of bonds, it may be challenging to find a suitable CDS (or CDS sector) curve. 

https://www.nomura.com/resources/europe/pdfs/cva-cross-section.pdf comes to mind and a number of papers that cite this one.

Also I wonder how you determine the CDS/bond basis (the arXiv paper picks up on this).

 

 

Best regards,

 

Matthias

 

From: Philippe Hatstadt <phi...@ex...> 
Sent: Thursday, 24 September 2020 18:56
To: QuantLib users <qua...@li...>
Subject: [Quantlib-users] Corporate bond analytics approach

 

I am looking to develop a quantitative infrastructure for corporate bonds. As I see it, there are two ways to proceed as below.

I am asking this group whether my summary of these two options is accurate or whether there are better approaches?

 

Approach 1: Simple Interpolated Spread Curve

Under this approach, bond cash flows would be discounted at UST + spread, from a curve built by shifting a QL UST curve by a bond specific corporate z-spread. In turn, such corporate z-spread would be interpolated from a z-spread curve, itself calibrated by a set of benchmark corporate bonds. 

 

Approach 2: CDS / Hazard Rate Curve

This approach would first use existing QL CDS helpers and hazard rate curves to calibrate such curve to observable CDS quotes. In turn, a corporate bond could be repriced by adding a "basis" spread on top of the CDS curve.

However, while QL appears to have a full set of methods and classes for CDS and hazard rate curves, it doesn't seem to have any analytics or methods to price corporate bonds on top of a CDS curve as described above

 

Respective Methods Pros and Cons:

All else equal, I would favor Approach 2 because it allows both bonds and credit derivatives to be priced and risk managed, as CDS curves are conveniently available on standardized IMM dates. However, that requires the layer of corporate bond analytics and classes on top of a hazard curve that I do not believe exists in QL.

 

I am probably missing something, so most curious about what QL users think about this particular problem? 

 

 




Philippe Hatstadt

 

 

 

Brokerage services offered through Exos Securities LLC, member of  <http://www.sipc.org/> SIPC /  <http://www.finra.org/> FINRA. For important disclosures,  <https://www.exosfinancial.com/disclosures> click here.